What you loosely describe could possibly be interpreted as a Generalized Estimating Equations model which can be fit in PROC GEE or PROC GENMOD. Another possibility, if the response is binary, is a conditional logistic model which can be fit using the STRATA statement in PROC LOGISTIC. You might want to look through the book "Fixed Effects Regression Methods for Longitudinal Data Using SAS"  (Allison, P., SAS Institute, 2005). ... View more

The statistics in your output data set are observation-wise statistics, which means you get one for each boservation. Perhaps you prefer to output model statistics, such as the SEB (standard errors of the parameter estimates) and the PRESS statistic? If so, look at the doc for the PROC REG statement and use the OUTEST= option, as follows:   data stocks; set sashelp.stocks(rename =(Stock=permno)); fyear = year(Date); run; proc sort data=stocks; by permno fyear; run; proc reg data=stocks noprint outest=Est OUTSEB PRESS; by permno fyear; model Close = Open; quit; proc print data=Est(obs=10); run; ... View more